04-28-2021 09:03 AM - edited 04-28-2021 09:57 AM
One of my customer has a 3-node production HXAF220C-M5SX HyperFlex cluster. Each node has 8 x 960GB SSD installed with ESXi and vCenter running. They now want to purchase replacement SSDs to increase the overall usable HX cluster storage. I never done this kinda replacement and I honestly do not think this can be done for a brownfield environment (but I could be wrong)...
I did went through the HX DP guide and hardware installation guide, but:
1. According to the HX DP guide, "Mixing storage disks type or size on a server or across the storage cluster is not supported.". This means I have to shut down the entire cluster (all three nodes) to replace the 8 drives each, right?
2. If so, how on earth would this kinda replacement still maintain the existing data on the cluster storage?
Please advise.
Solved! Go to Solution.
04-28-2021 11:15 AM
The cluster will need to be rebuilt with the larger capacity drives in place.
You would need to have some additional storage outside of hyperflex, in order to copy/move the guestVMs off to prior to the cluster rebuild.
Kirk...
04-28-2021 11:16 AM
To summarize the ask.
You want to remove existing smaller disks (say 500GB) and replace them with new larger disks (say 2TB)?
Yeah, can't do that "live". Would be a backup, wipe, redeploy.
Could maybe be done technically, but this isn't a supported/tested/qualified workflow to accomplish this.
Would take more Engineering work (if it even worked at all) compared to simply redeploy. Sorry.
04-28-2021 11:15 AM
The cluster will need to be rebuilt with the larger capacity drives in place.
You would need to have some additional storage outside of hyperflex, in order to copy/move the guestVMs off to prior to the cluster rebuild.
Kirk...
04-28-2021 11:16 AM
To summarize the ask.
You want to remove existing smaller disks (say 500GB) and replace them with new larger disks (say 2TB)?
Yeah, can't do that "live". Would be a backup, wipe, redeploy.
Could maybe be done technically, but this isn't a supported/tested/qualified workflow to accomplish this.
Would take more Engineering work (if it even worked at all) compared to simply redeploy. Sorry.
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide